HEXIS · SETTLED STATE
On the difference between an incident and a disposition — and on the decades-long argument about whether dispositions exist.
Somebody has a good week. Everything they meant to do, they did. By Friday there is a sentence forming: I have changed.
They have not, or at least the week is not evidence that they have. And the same person after one bad week produces the opposite sentence with equal confidence, which is the tell: if a single instance can establish it either way, it was never an inference about a disposition.
Aristotle’s term is hexis, usually rendered as a settled state or disposition. Not a mood and not a resolution — a condition of a person that makes a certain action the one they take without particularly deciding to.
One failure over-reads a single week. The other refuses to believe anything is being built.
One instance taken as evidence of a disposition. It feels like progress and it is a sampling error, and the same error runs in reverse on a bad day and costs considerably more.
“There is no such thing as character, only situations.” There is a serious version of this argument, below. Taken as a conclusion it removes any reason to repeat anything, which is not what the evidence supports.
He uses it twice, and this collection ends on its second appearance. Here it is about evidence; in essay twelve it is about assessment.
The claim is about what kind of thing a disposition is. It is a summary over many instances, so no single instance can establish or refute it — which is inconvenient for anybody who wants to know how they are doing this week.
Aristotle’s illustration is a builder. You become one by building, and there is no other route — not by studying building, not by intending to build, not by having the aptitude.
He then meets the obvious objection head on, which is one of the better moments in the book. If we become just by doing just acts, must we not already be just in order to do them?
His answer is that the early acts are not done the way the settled person does them. They are done deliberately, awkwardly, on instruction, without the state behind them. That is what apprenticeship is, and it is not circular because the two kinds of act differ in exactly the way that matters.
William Fleeson’s work changed the shape of this argument.4 Rather than treating a trait as a fixed level, he had people report their behaviour repeatedly over days and found that each person produces a wide distribution of behaviour — and that the distribution’s central tendency is stable even though any single sample is nearly uninformative.
The structure of that finding resembles Aristotle’s claim, and it is worth being exact about what it does and does not do. It establishes that behaviour sampled once is a poor guide to a person’s central tendency — which is the practical half of what he says. It establishes nothing about whether the stable centre is a hexis in his sense, whether it was built by repetition, or whether it is the kind of thing a person is responsible for. Those are his claims and Fleeson did not test them.
What you get, then, is one usable fact: a person has a characteristic distribution, and a single Tuesday tells you almost nothing about it.
It also makes the practical point exactly: you cannot read your own disposition off a week, because a week is a small sample from a wide distribution. What you can observe is the density of instances over months.
The person-situation debate is the strongest objection here, and it was not settled in Aristotle’s favour.
The case against
Walter Mischel’s critique showed that cross-situational consistency in behaviour is much lower than trait language implies, and that situations predict a great deal of what people do. Decades of work followed and the modern position is an interactionist one: dispositions are real and weaker than intuition says, and situations do a lot of the work.
The practical consequence is not that character is a fiction. It is that arranging the situation is often a better investment than building the disposition, and that a person who does the right thing in an environment built for it has not necessarily demonstrated anything about themselves.
This is also where character talk does harm. A disposition is stable by definition, so reading a failure as evidence about your character converts a bad month into a life sentence. The honest version is that you have a distribution and you are trying to move its middle, slowly.
Claim nothing from a week. Not to yourself. A good week is a good week, and the sentence “I have changed” is not available on that evidence.
Count instances instead. Over a month, how many times did the thing happen? That is a sample worth something, and it is the only kind of evidence the category admits.
Watch the reverse move especially. A bad week produces “I am the kind of person who” far faster than a good one, and it costs more, because the sentence changes what you attempt next.
And arrange the situation, given the section above. Most of what looks like discipline is an environment in which the thing is easy, which is not cheating — it is the better-supported half of the evidence.
The first job is noticing how much of your self-description rests on single instances.
You become a builder by building. There is no faster route, and the thing you become is not a fact about a week.
What the modern evidence adds is that the distribution is wide and the situation matters more than the language of character suggests. Both halves are useful: repetition does build something, and arranging the environment does more than intending harder.
The one sentence to retire is “I am the kind of person who.” On a good week it is premature. On a bad one it is expensive.
A disposition is a summary over many instances, so no single week establishes or refutes one — and the reverse move, after a bad week, is the expensive one.
The person-situation debate did not go Aristotle’s way. Dispositions are real and weaker than intuition says, and arranging the environment is usually the better investment.
Quotations are from the Nicomachean Ethics unless noted, cited by book, chapter and Bekker number; renderings follow W. D. Ross (rev. Urmson), Terence Irwin or Roger Crisp, and a condensed line is marked “after Aristotle”. Modern research is cited by author and publication.